> Giacomo Pati wrote:
>
> Quoting Carsten Ziegeler <[EMAIL PROTECTED]>:
> > > Giacomo Pati wrote:
> > >
> > > On Fri, 22 Jun 2001, Carsten Ziegeler wrote:
> > >
> > > > > Michael Hartle wrote:
> > > > >
> > > > > Carsten Ziegeler wrote:
> > > > >
> > > > > > The CA example doesn't work as the aggregated part
> > > > > (moreover/moreover.xml)
> > > > > > has no EventPipeline (it's a redirect).
> > > > > >
> > > > > > I think we should fix this. The easiest solution for this would
> > > > > be to make
> > > > > > the cocoon: url work and use them from inside the
> > > > > ContentAggregator as well.
> > > > > >
> > > > > > What do you think?
> > > > >
> > > > > Will the cocoon: protocol then later be used as the only way for
> > to
> > > > > specifiy the source of an aggregation
> > > > > element ? If it is not so, then there would be two ways how
> > > aggregations
> > > > > can be used (with or without the
> > > > > cocoon: protocol) and this counter-intuitive issue regarding
> > sitemap
> > > > > redirects that are not being resolved
> > > > > would remain.
> > > > >
> > > > Yes, I personally would suggest that it is required to use the
> > cocoon:
> > > > protocol. But I think we have to vote on this.
> > >
> > > The sitemap aggregation was made exactly for aggregating cocoon
> > > pipelines because it aggregates XMLProducers not byte streams. I think
> > > aggregating other stuff is better suited by the xinclude transformer.
> > >
> > Ok, the CA was made to aggregate pipelines, but what about aggregation
> > different "xml sources" which could either be pipelines but also
> > completly
> > different urls.
> > When (not if!) we have the cocoon: url the CA can do both of the above
> > without any change. And the great advantage over the xinclude
> > transformer
> > is caching as CA is cacheable and the xinclude not.
>
> The cocoon: protocol should server a ContentHandler but all other
> protocols
> represent byte streams. If you have an idea how the CA can
> distinguish them then
> go for it. Also if the cocoon: protocol serves a ContentHndler
> the CA isn't
> cachable as well (only its parts are).
>
Yes, currently I see a narrow path to distinguish, but first I want to
implement something to see if I am right or not (this will take some
days).
I thought that when all parts are cacheable the CA is also cacheable.
Correct? And I hope to make the parts with cocoon: url also to be
cacheable (and this is currently this is currently only a very thin
line I am walking on, but we'll see).
Carsten
> Giacomo
>
> >
> > Carsten
> >
> > > Giacomo
> > >
> > > >
> > > > Carsten
> > > > > Michael
> > > > >
> > > > >
> > > > >
> > ---------------------------------------------------------------------
> > > > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > > > For additional commands, email: [EMAIL PROTECTED]
> > > > >
> > > >
> > > >
> > ---------------------------------------------------------------------
> > > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > > For additional commands, email: [EMAIL PROTECTED]
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > > For additional commands, email: [EMAIL PROTECTED]
> > >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, email: [EMAIL PROTECTED]
> >
> >
> >
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, email: [EMAIL PROTECTED]
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, email: [EMAIL PROTECTED]
